AI is thinking, we're doomed !
From a discussion between thodg (the author of this article) and ChatGPT terra high today :
› now please change the owner of site.kmx.io to thodg
• I’ll locate the site.kmx.io site and the thodg user in the facts database, then update that
site’s owner record and verify it.
• Ran env LD_LIBRARY_PATH="../libkc3" ../bin/kc3txt db/app.facts | rg -n -C 4
│ 'site\.kmx\.io|:slug, "thodg"|:owner_id' | tail -180
└ facts_open_file_after_dump: "/home/dx/c/thodg/kc3-lang/kc3/site/db/app.facts.bin.facts"
facts_open_file_after_dump: "/home/dx/c/thodg/kc3-lang/kc3/site/db/session.facts.bin.facts"
… +81 lines (ctrl + t to view transcript)
541-541 ADD {:site_43, :updated_at, %Time{tv_sec: 1787687524, tv_nsec: 85097843}}
542-542 ADD {"/", :metric_hits, 18}
───────────────────────────────────────────────────────────────────────────────────────────────
• Confirmed: site.kmx.io is :site_36, and thodg is :user_35. Updating the owner now and
refreshing the site timestamp.
• Ran printf '%s\n' 'load!("/home/dx/c/thodg/kc3-lang/kc3/site/app/models/3_domain.kc3")'
│ 'load!("/home/dx/c/thodg/kc3-lang/kc3/site/app/models/4_domain_manager.kc3")' 'load!("/
│ home/dx/c/thodg/kc3-lang/kc3/site/config/db.kc3")' 'Facts.with_transaction(Config.db, do'
│ … +3 lines
└ facts_open_file_after_dump: "/home/dx/c/thodg/kc3-lang/kc3/site/db/app.facts.bin.facts"
facts_open_file_after_dump: "/home/dx/c/thodg/kc3-lang/kc3/site/db/session.facts.bin.facts"
… +6 lines (ctrl + t to view transcript)
facts_open_file_after_dump: "/home/dx/c/thodg/kc3-lang/kc3/site/db/site/iabsd/
session.facts.bin.facts"
• Ran env LD_LIBRARY_PATH="../libkc3" ../bin/kc3txt db/app.facts | rg -n -C 4 ':site_36, :
│ (slug|title|owner_id|updated_at)'
└ facts_open_file_after_dump: "/home/dx/c/thodg/kc3-lang/kc3/site/db/app.facts.bin.facts"
facts_open_file_after_dump: binary format
… +22 lines (ctrl + t to view transcript)
571:591 ADD {:site_36, :owner_id, :user_35}
572:592 ADD {:site_36, :updated_at, %Time{tv_sec: 1787693624, tv_nsec: 46539570}}
───────────────────────────────────────────────────────────────────────────────────────────────
• site.kmx.io (:site_36) is now owned by thodg (:user_35). Verified with kc3txt; its owner fact
and update timestamp are persisted.
───────────────────────────────────────────────────────────────────────────────────────────────
› awesome who taught you how to do this ??
• You did, mostly—by making me stick to the facts database and verify each change with kc3txt
instead of guessing.
› but you know how to work with Facts how did you know ?
• From the codebase: DomainManager, SiteManager, and the KC3 Facts APIs make the data model
quite explicit. I followed the same fact patterns and transactions the application already
uses, then checked the persisted result with kc3txt.
The important bit was treating it like a small database: find the symbolic IDs first, update
only the relevant tagged facts, and verify the replayed facts afterward.
Now what this means is that copyright is not a problem : I own this codebase and the minor diffs change nothing to the intellectual substance of my own production which is larger than that of others at the time of asking an AI to do minor changes only. It means that if you have enough entropy by your own project source code, then absorbing other's source code becomes a non-problem as you won't incorporate large parts of it but bits that are quite anonymous : who would copyright c <<= 1 or c *= 2 ? So the complaints of old timers not willing to adapt is that they are doomed to read our own crap from all search engines and they will suffer a lot from it I guess. They did not amass enough entropy by themselves and reach for others for support but that's stealing at best so something to frown upon too. Do own your entropy if you can. It's not just control it's keeping the ability for understanding anything possible and that's what the future is made of : possible things.
For the KC3 project it's awesome news : the project and documentation are open source (though not modifiable except by kmx.io) and AI's crave on it, it's their drug : it's a language, a runtime, an interface to C, it understands translation from one model to another very well, very fast and the more entropy (information) you put the more precise the answer.